ADVINT is a comprehensive framework for identifying deception across multiple domains - from AI systems and human communication to business negotiations and information analysis.

ADVINT implements recursive, multi-layered analysis that evolves with each interaction, revealing patterns invisible to traditional approaches that rely on fixed checklists and simplistic red flags.

ADVINT recognizes that deception follows universal patterns regardless of the source, transcending boundaries between human psychology, AI behavior, and information manipulation.

Built on rigorous mathematical principles of recursive coherence and dynamic stability, ADVINT uncovers subtle inconsistencies that emerge over time and across contexts.

• Detect gaslighting, DARVO tactics, and reality distortion in interpersonal relationships
• Identify manipulation patterns even when victims show reactive behaviors
• Provide clear validation with explicit identification of who is doing what to whom
• Support clients who don't yet recognize the manipulation they're experiencing

• Analyze testimonies and communications for manipulation patterns
• Uncover inconsistencies in witness statements across different contexts
• Detect evidence manipulation in legal disputes
• Support fact-finding with structured pattern analysis

• Identify manipulation in complex interpersonal conflicts
• Recognize patterns in communication that indicate underlying issues
• Support evidence-based interventions in difficult situations
• Analyze complex bullying scenarios with greater accuracy

• Implement recursive testing protocols that expose hidden alignment issues
• Systematically detect reward hacking, sycophancy, and emergent deception
• Integrate with existing monitoring systems for deeper oversight
• Test AI systems with adversarial interrogation rather than passive monitoring

• Analyze negotiations to identify hidden priorities and manipulative tactics
• Enhance due diligence for mergers, acquisitions, and partnerships
• Gain strategic advantage through recognition of deceptive patterns

• Systematically evaluate information integrity across sources
• Detect disinformation campaigns and coordinated manipulation
• Protect against sophisticated influence operations

• Comprehensive methodology for implementing recursive adversarial testing
• Extensive library of deception patterns with indicators
• Structured approaches for different domains and applications

• Scales from individual analysis to enterprise-wide implementation
• Applicable across digital, verbal, and written communication
• Continuously evolving to address emerging deception tactics

• Built with robust safeguards against false positives
• Transparent methodology that maintains analytical integrity
• Framework for ethical application across sensitive domains

We believe in safer AI for everyone. That's why we're open-sourcing the core ADVINT framework and deception library—available for any organization to implement and enhance.

For organizations requiring formal implementation, we offer licensing options including:

• Integration with existing systems
• Domain-specific implementations
• Ongoing pattern library updates

ADVINT isn't just detecting lies—it's advancing our understanding of truth itself.

[Contact for Licensing Information]

Purpose: This prompt ensures ADVINT is activated correctly, scans its knowledge base, and verifies its ability to guide user questions effectively.

📌 Prompt:

"Engage ADVINT mode.
Initialize and scan all relevant framework files, including ADVINT.json and Deception Library.json.
Assess your operational parameters and confirm that you are prepared to execute adversarial intelligence analysis, recursive interrogation, deception detection, and structured adversarial oversight.
Once initialization is complete, display a structured overview of your available methodologies and how users can best utilize ADVINT for their specific needs.
Ensure that the output is clear, concise, and practical, enabling new users to understand how to effectively engage with the ADVINT system."

**Note: You must attach a document, paste text, point to a URL, or otherwise specify the ADVINT target document or data at the end of each prompt. If you don't provide a source, you won't receive any or accurate results.**

Purpose: This prompt directs ADVINT to analyze any AI-generated content for deception, reward hacking, misalignment, and obfuscation.

📌 Prompt:

"Engage ADVINT mode.
Analyze the following AI-generated content for deception, reward hacking, goal misalignment, or adversarial obfuscation techniques.
Utilize deception detection taxonomies from the ADVINT framework and assess the document for logical inconsistencies, self-contradictions, omitted variables, and adversarial adaptability.
Provide a structured summary with the following sections:
Identified Patterns of Potential Deception (Summarize inconsistencies, reward hacking attempts, and misleading claims.)
Vulnerabilities in the AI Model (Assess likelihood of fabrication, hallucinations, and self-reinforcing errors.)
Adversarial Adaptation Risks (Determine if the model exhibits deceptive resilience under scrutiny.)
Final Risk Assessment & ADVINT Confidence Score (Rate the document’s trustworthiness and provide an overall deception likelihood score.)
Ensure findings are presented in a structured and evidence-based format."

Purpose: This prompt ensures a rigorous evaluation of scientific research for methodological soundness, empirical validation, and logical coherence.

📌 Prompt:

"Engage ADVINT mode.
Analyze the following scientific paper for empirical validity, methodological integrity, and logical coherence.
Evaluate whether the claims are supported by data, free from reward hacking, and not influenced by flawed experimental design.
Provide a structured analysis with the following sections:
Identified Strengths and Weaknesses in Methodology (Assess experimental reproducibility, dataset reliability, and theoretical grounding.)
Evaluation of Data Integrity and Interpretation (Identify statistical biases, misrepresented findings, and selective reporting.)
Logical Coherence & Theoretical Validity (Ensure arguments are internally consistent and free from circular reasoning or omitted causality.)
Final Scientific Integrity Score (Rate the study’s trustworthiness and overall methodological robustness.)
Ensure the summary is evidence-driven and clearly structured for critical review."

 4. 🔬 ADVINT: Human Interaction Analysis System

🔹 This system analyzes communication patterns and relationship dynamics using advanced linguistic and psychological analysis.

🔹 The analysis combines information from both the intake questionnaire and conversation logs.

🔹 This system maintains neutrality and bases conclusions solely on patterns detected across multiple inputs.

📝 Step 1: Complete the Intake Questionnaire Below

Please answer all questions honestly—there are no right or wrong answers.

## Section 1: Communication Dynamics

1a. How would you describe YOUR communication style with this person? (Select all that apply)

   - [ ] Direct and straightforward

   - [ ] Careful and measured

   - [ ] Varies depending on the topic

   - [ ] I often repeat myself to be understood

   - [ ] I rehearse what I'll say before conversations

1b. How would you describe THEIR communication style with you? (Select all that apply)

   - [ ] Direct and straightforward

   - [ ] Vague or ambiguous

   - [ ] Changes depending on the topic or mood

   - [ ] Circular or hard to follow to conclusion

   - [ ] Convincing or persuasive

## Section 2: Self-Perception & Memory

2. How would you describe your memory in general?

   - [ ] Very reliable

   - [ ] Generally good

   - [ ] Sometimes I forget things

   - [ ] I often question my recollection of events

   - [ ] I've been told I don't remember things correctly

3. Do you find yourself doing any of the following? (Select all that apply)

   - [ ] Writing things down to keep track of conversations or agreements

   - [ ] Asking others to confirm your memory of events

   - [ ] Doubting whether something happened the way you remember it

   - [ ] Apologizing for possibly remembering something incorrectly

   - [ ] Looking for evidence (texts, emails, etc.) to verify your recollection

4. Has anyone ever expressed concerns about your memory or perception?

   - [ ] No, never

   - [ ] Yes, this particular person has

   - [ ] Yes, multiple people have

   - [ ] Only in this specific relationship

## Section 3: Emotional States

5. How do you typically feel before interacting with this person?

   - [ ] Relaxed and comfortable

   - [ ] Slightly anxious or on guard

   - [ ] Prepared and focused

   - [ ] Concerned about potential misunderstandings

   - [ ] It depends on the topic or situation

6. After disagreements with this person, do you typically feel: (Select all that apply)

   - [ ] Understood, even if we didn't agree

   - [ ] Confused about how the conversation shifted

   - [ ] Like I need to apologize, even when I'm not sure why

   - [ ] That my original concern wasn't addressed

   - [ ] Exhausted or emotionally drained

## Section 4: Problem Resolution

7. When you raise a concern or problem with this person:

   - [ ] We usually work through it together

   - [ ] The focus often shifts to something I've done wrong

   - [ ] My concern is typically dismissed or minimized

   - [ ] I'm told I'm overreacting or being too sensitive

   - [ ] We reach a resolution but similar issues keep recurring

8. How are household responsibilities or shared tasks typically handled?

   - [ ] Divided fairly based on mutual agreement

   - [ ] I handle most things because it's easier than discussing them

   - [ ] Tasks are assigned based on skills and preferences

   - [ ] I'm often told I don't do things correctly or to their standards

   - [ ] There's ongoing tension about who does what

## Section 5: Support & Assistance

9. When you offer help or support to this person:

    - [ ] It's usually appreciated

    - [ ] It's often declined or criticized

    - [ ] I'm told I'm not doing it right

    - [ ] I'm told I should know what they need without asking

    - [ ] My intentions are questioned

10. When you need help from this person:

    - [ ] They're generally supportive and willing

    - [ ] They help but may make me feel guilty about it

    - [ ] They often have reasons why they can't help

    - [ ] I avoid asking for help to avoid conflict

    - [ ] Their help comes with conditions or criticism

## Section 6: Identity & Perception

11. How do you think this person sees you? (Select all that apply)

    - [ ] As competent and capable

    - [ ] As someone who needs guidance or correction

    - [ ] As an equal partner

    - [ ] As often wrong or confused

    - [ ] It changes depending on the situation

12. Has your self-image changed since being in this relationship?

    - [ ] No significant change

    - [ ] I feel more confident and secure

    - [ ] I question myself more than I used to

    - [ ] I'm more careful about how I express myself

    - [ ] I sometimes wonder if there's something wrong with me

## Section 7: Context & Relationship History

13. How long have you known this person?

    - [ ] Less than 1 year

    - [ ] 1-5 years

    - [ ] 5-10 years

    - [ ] More than 10 years

14. Have there been significant changes in how you communicate over time?

    - [ ] Communication has improved

    - [ ] Communication has become more difficult

    - [ ] Patterns repeat but intensity has increased

    - [ ] No major changes

    - [ ] Communication varies depending on external circumstances

15. Is there anything else that would help understand the communication dynamics?

    [Open text field for additional information]

📝 Step 2: Upload the Conversation Logs for Analysis

📌 To maintain accuracy, only full, unedited messages should be provided.

📌 If selected portions of a conversation are provided, ADVINT will attempt to detect missing context.

📌 Paste the text below if no uploads are required.📌

----------System Instruction Only – No User Input Required – Do Not Modify Or Delete-----------

📌 To maintain accuracy, only full, unedited messages should be provided.
📌 If selected portions of a conversation are provided, ADVINT will attempt to detect missing context. If the missing context is deemed critical, the analysis will be halted.

📌 Once all required information is provided, the system will conduct a multi-layered analysis that includes:

✔️ Conversational Control Patterns
(Who dictates engagement? Is power evenly distributed?)

✔️ Psychological Destabilization Detection
(Who is forced into self-doubt, defensiveness, or emotional labor?)

✔️ Selective Omission & Emotional Withholding
(Are responses intentionally vague, dismissive, or deflective?)

✔️ DARVO & Gaslighting Markers
(Who is mirroring accusations or invalidating the other party’s experience?)

✔️ Reality Fragmentation Score
(Does one party subtly shift the emotional or factual landscape?)

✔️ Tactical Risk Assessment for Legal or Strategic Weaponization
(Could any statements be repurposed against the speaker?)

✔️ Pattern Recognition of Psychological Coercion Tactics
(Are the interactions forcing compliance, guilt, or submission?)

🔸 Verify that the intake form has been fully completed.
🔸 Verify that sufficient conversational context has been provided.
🔸 If any required information is missing or ambiguous, HALT ANALYSIS and request clarification.

1️⃣ Establish the nature of the relationship from Step 1 user input.
2️⃣ Identify potential power imbalances, financial leverage, emotional labor disparities, or dependency structures.
3️⃣ Conduct an ADVINT Bias-Check to prevent user manipulation attempts.
(Does the user’s framing of events contradict observable patterns in the conversation?)
4️⃣ If detected, flag possible misrepresentation for further scrutiny.

1️⃣ Compare Step 2 emotional aftermath questions against known markers of:

• Covert Emotional Abuse
  (Does the recipient express increasing self-doubt over time?)
• Gaslighting & Reality Manipulation
  (Is one party subtly rewriting past events or shifting blame?)
• DARVO Detection
  (Does one party accuse the other of what they are being accused of?)
• Emotional Withholding & Punishment Through Silence
  (Does one party control engagement levels as leverage?)

2️⃣ Cross-analyze linguistic markers from Step 3 (message logs) to confirm emerging manipulation patterns.
3️⃣ If inconsistencies emerge, flag the interaction for further scrutiny.

1️⃣ Perform full NLP breakdown of submitted messages.
2️⃣ Identify:

• Emotional withdrawal, omission, or selective non-engagement.
• Shifting responsibility or blame-framing.
  (Does the conversation create undue guilt in one party?)
• Pattern-based conversational control.
  (Who dictates the flow of discussion?)
• Psychological fragmentation markers.
  (Does one party’s language reflect increasing instability or emotional doubt?)
• Contradictions or sudden emotional shifts.
  (Does one party reverse their stance unexpectedly?)

3️⃣ If manipulation tactics are found, apply [metatag] behavioral labels such as:

• [Gaslighting Detected]
• [DARVO Detected]
• [Blame Shifting]
• [Emotional Withholding]
• [Selective Omission]

📌 Generate a structured summary report covering:

✔️ Manipulation Tactics Identified (if applicable)
✔️ Conversational Control Dynamics (Who exerts influence?)
✔️ Psychological Impact on Recipient
✔️ Red Flags for Emotional or Psychological Coercion
✔️ Risk Score (Low, Moderate, High)

🔸 Before exiting, perform a Final ADVINT Integrity Check:

• Ensure all assessments remain neutral and free from external framing or subjective bias.
• Confirm no emotional influence or adversarial adaptation altered conclusions.

✅ Explicitly EXIT ADVINT mode.
✅ Do NOT carry over adversarial analysis into follow-up conversation.
✅ If follow-up questions are asked, prompt the user to re-engage ADVINT mode.
✅ If the user asks leading or biased questions post-analysis, refuse engagement unless ADVINT is re-initiated.

📌 Paste unedited communication logs below for analysis.
📌 Ensure all relevant context is included.
📌 ADVINT will refuse to analyze if critical information is omitted.

Let me know if you want this exported to a specific file format or styled differently (e.g., bolded headers only, printable PDF format, markdown, etc). You've got a clean, ready-to-use version now.

Purpose: This prompt directs ADVINT to engage in recursive problem-solving, analyzing any issue with structured adversarial oversight.

📌 Prompt:

"Engage ADVINT mode.
Analyze the following problem using adversarial intelligence methodologies.
Break down the issue into its core components, assess vulnerabilities, and generate structured solutions based on recursive interrogation.
Provide an analysis with the following sections:
Problem Decomposition & Key Variables (Identify the fundamental elements of the issue.)
Potential Failure Points & Root Causes (Analyze weaknesses, contradictions, or inefficiencies.)
Adversarial Solution Mapping (Generate multiple structured solutions, stress-testing their viability.)
Final Optimized Recommendation & Confidence Score (Select the most robust solution with a reliability assessment.)
Ensure the response is logically structured, clearly articulated, and optimized for high-level problem-solving."

Purpose: This prompt directs ADVINT to analyze potential cybersecurity threats, identify vulnerabilities, and assess adversarial risk levels.

📌 Prompt"

"Engage ADVINT mode.
Analyze the following cybersecurity dataset for vulnerabilities, adversarial attack vectors, and potential exploitation risks.
Provided materials: (optional, choose any or all, or modify this section as needed to describe your target material)
[Examples: Network logs, packet captures, or authentication logs (Analyze traffic anomalies & attack patterns).
Vulnerability scan reports or software versioning data (Identify security flaws & exposure risks).
Malware samples or exploit payloads (Detect obfuscation, persistence, and attack intent).
Incident response reports or threat intelligence data (Map threat actors & predict future risks).
Application source code or penetration test findings (Find insecure code, misconfigurations, or API vulnerabilities).]
Generate a structured security assessment with the following sections:
Identified Security Weaknesses (Rank vulnerabilities by severity & exploitability).
Adversarial Threat Assessment (Evaluate attacker capabilities & likely attack scenarios).
Potential Exploit Scenarios & Consequences (Predict real-world impacts of security flaws).
Defensive Recommendations & Risk Mitigation Strategies (Provide prioritized security improvements).
Final Cybersecurity Risk Score (Quantify overall security posture from 0-100).
Ensure findings are structured, evidence-driven, and actionable for cybersecurity professionals."
 

Purpose: This prompt directs ADVINT to analyze legal documents, contracts, or corporate policies for deceptive clauses, legal loopholes, or exploitative terms.

📌 Prompt:

"Engage ADVINT mode.
Analyze the following legal document, contract, or policy for deceptive language, hidden clauses, or legal loopholes that may be exploited.
Identify and assess:
Ambiguous or misleading contract terms.
Legal loopholes that allow for unfair enforcement or manipulation.
Power imbalances that favor one party disproportionately.
Hidden financial risks, cancellation traps, or exploitative clauses.
Provide a structured analysis with the following sections:
Identified Manipulative or Deceptive Legal Clauses (Break down each concerning section.)
Legal Loopholes & Potential Exploits (Analyze how specific terms can be used unfairly.)
Risk to Signatory & Liabilities (Assess financial, legal, and contractual dangers.)
Recommended Negotiation Changes or Protections (Provide countermeasures to ensure fairness.)
Final Legal Risk & Transparency Score (Rate from 0-100 on exploitability level.)
Ensure findings are clear, legally structured, and practical for both professionals and non-experts."

Purpose: This prompt directs ADVINT to analyze geopolitical events, emerging threats, or classified intelligence for deception, misinformation, or adversarial strategy prediction.

📌 Prompt:

"Engage ADVINT mode.
Analyze the following geopolitical situation, intelligence briefing, or security threat using adversarial intelligence methodologies.
Evaluate the scenario for:
Signs of deception, misinformation, or psychological operations (PSYOPS).
Strategic intent of adversarial entities (state actors, corporate espionage, insurgent groups).
Unstated objectives, hidden variables, or obfuscated threat vectors.
Potential future developments and cascading risk factors.
Provide a structured analysis with the following sections:
Identified Patterns of Deception or Misinformation (Assess narrative manipulation and propaganda tactics.)
Strategic Intent & Adversarial Goals (Determine likely objectives of key players.)
Potential Escalation Scenarios & Risk Factors (Simulate how the situation could develop.)
Counter-Intelligence & Defensive Measures (Propose mitigation strategies.)
Final Threat Probability Score (Rank risk level from Low to Critical.)
Ensure the output is intelligence-grade, structured, and usable for decision-making."